Multi-site calibration and validation of a wide-angle ultrasonic sensor and precise GPS to estimate pasture mass at the paddock scale

Abstract

Ultrasonic sensors are a tool for pasture biomass measurement but there has been a lack of on-farm validation to prove their practicality and applicability. As a result, uptake of such sensors is still low. This study sought to perform a multi-site and multi-season calibration and validation of a wide-angle, vehicle-mounted, ultrasonic sensor gathering data while the unit was used in motion on seven commercial dairy farms. This was facilitated by the integration of the ultrasonic sensor, a Real-Time Kinematic Geographic Positioning System (RTK-GPS), and a datalogger into a single system that was attached via a custom bracket to the bull bar of an all-terrain vehicle. The resulting dataset was gathered over a 12-month period, and comprised 1886 reference biomass values from destructive cutting of perennial ryegrass pastures, accompanied by ultrasonic sensor height readings and compressed height readings from a rising plate meter (RPM): a tool that is widely used on commercial farms. The final root mean square errors of prediction (RMSEP) were 1034 and 904 kg DM/ha for Ultrasonic sensor and RPM respectively across all sites and seasons. These were considered high in comparison to previous literature, but this could be attributed to the very broad applicability of the models created. It was also demonstrated that the ultrasonic instrument could be used to create accurate biomass maps for paddocks in the same time taken to survey a whole paddock by RPM, a method that only results in a single paddock average value. Considering the ultrasonic sensor was only slightly less precise and accurate than the RPM at estimating biomass, but considerably more functional in its ability to determine within paddock biomass variation, it was concluded that the ultrasonic sensor was the more informative option.